Unique SMILES generation changes after calling getChirality

We've found a very strange problem with unique SMILES generation. This is somewhat related to a previous post:


http://chemaxon.com/forum/ftopic2434.html





What we discovered is that unique SMILES generated for some compounds can be different if Molecule.getChirality() is called anywhere on a chiral atom -- even in a completely different molecule!





Here's some sample code. I tested this with the latest 4.1.6 version released today. In this example, we have a SMILES that should be unique.


Importing it into a molecule and re-exporting as unique SMILES produces the same SMILES, as expected -- unless you happen to call Molecule.getChirality first.

Code:

// Getting the chirality for a chiral atom seems to affect the       // behavior of the unique SMILES generation       Molecule dummyMolecule = MolImporter.importMol("[C@H](N)(S)(O)");       dummyMolecule.getChirality(0);      // Removing this line (or changing "0" to "1") makes the code below work!       // This is the original SMILES that should be unique       String originalSmiles = "[CH3:2][CH2:3][C:4](=[O:5])[NH:6][CH2:7][CH:8]1[CH2:9][CH2:10][N:11]([CH2:12]1)[C:33]2=[C:13]([C:29]3=[C:30]([C:34](=[O:35])[C:26](=[CH:27][N:28]3[CH2:19][CH:20]4[CH2:21][CH2:22]4)[C:24]([OH:23])=[O:25])[C:31]([NH2:18])=[C:32]2[F:1])[C:14]([F:15])([F:16])[F:17]";       Molecule mol = MolImporter.importMol(originalSmiles);              // Make it unique (again).  We expect this to be the same as the original SMILES       // above, but the call to getChirality(0) above changes the behavior.       String uniqueSmiles = mol.toFormat("smiles:u-a");       assertEquals(originalSmiles, uniqueSmiles);  // Fails unless the getChirality line above is removed

There are a few obvious questions:





- How could calling Molecule.getChirality on an unrelated molecule affect the unique SMILES generation? There must be some kind of global state that is initialized or changed when calling getChirality on a chiral atom.





- Are there any other methods besides getChirality that could have similar effects?





- Can you provide a workaround or fix for this problem? Consistent generation of unique SMILES is absolutely critical for our product. (And we can't avoid calling getChirality.)





Thanks,


Chris

Thank you for the report, we are still examining the problem.

Seems managed to fix it, will ask Tamas to create a 4.1.7 pre-release of Marvin.

Hello, you can find the new Marvin and JChem at





http://www.chemaxon.com/test/marvin/index.html


http://www.chemaxon.com/download.php?d=/data/download/jchem/test





hoping it is all right, please give a try.





Szilva